Please YES more uncut Claudeslop Jev demos yes PLEASE don't think about it just UNGH just SLAP IT ON THERE just SLAP IT ON THERE, just ungh DEMO, DEMO, don't THINK, don't THINK ABOUT IT, just fucking DEMO, DEMO, DEMO, just DEPLOY TO THE CITY, CITY NEEDS IT, YES, YES, YES, YES
Raining in Chicago. Not one person offered to share their umbrella with me. Maybe it has to do with the fact that it’s all my fault, they haven’t forgiven me. They might never forgive me
I keep saying this but if you work at OpenAI and have concerns about AI risk you need to understand that your company’s superpac is one of the major proximate barriers to doing anything.
As much as I'd like this to work, 1-bit / ternary models like Bonsai are (today) a dud.
The sweet spot appears to be ~4 to 5 bits per parameter. That's what @UnslothAI find to be the "knee" of the pareto curve (thanks @realGeorgeHotz for pointing this out).
Interestingly a synapse in the human brain uses about 4.7 bits of information so perhaps this points to something more fundamental.
You're much better off using a smaller model quantized to 4-bit than a larger model quantized to 1-bit or ternary.
We also find these 1-bit / ternary models to be overly verbose - so they can actually run *slower* on end to end tasks than 4-bit, even though the decode TPS is faster.
Things have been moving quite fast in the past few weeks especially. It's encouraging how robust the discussion around new AI research has become. This, Jev, Bend 2, are exciting ideas and attempts, but I'm grateful the community does not blindly accept or praise them.
98.2% of Qwen3.8 27B my ass.
I got hyped and gave it a real agent job right away. Build me an FPS in three.js, 6 hours on a 3090. My most standard and default prompt that I always use. It spent the first 32K tokens on a plan without writing a single file, then shipped a black screen, 2 shaders that dont compile and a player who spawns dead. And wrote "verified" in the final report.
Ok, too hard. I gave it the easiest thing I have, a voxel pagoda garden in one html file. Video attached. 3 hours for THIS. On the way it deleted its own file and spent an hour debugging a raycaster nobody asked for.
Where the 98.2% is in all this I have no fucking idea. Same Qwen3.8 27B, same pagoda task, ISTA-DASLab GSQ-RCO IQ2_XS on a 12 GB 3080 Ti gave me day and night, real shadows and koi fish in the pond. 8.4 GB on disk, real 2.50 bpw, 131072 ctx with q4_0 KV, 47 tok/s at 128K. 2.5 bits beats 2.13 bits by a lot when the 2.13 is this shit. Post with the video and the exact command in the replies.
In the replies, I'll attach what a proper Qwen3.8 27B created in my hands. Bottom line: if you have at least 12GB VRAM, use ISTA-DASLab GSQ-RCO of all THIS.
Anthropic has set up a wet lab in San Francisco, their bio research has advanced beyond in silico. They are now conducting physical research both in-house and with external partners. The NYT also broke the news this morning that Anth now plans to scale its compute to 5 GW by the end of the year, and 10 GW within the next 16 months. Bio is next after math, and I think we are going to see some medical announcements much sooner than people are expecting.
@yishan Does saying something like "real atoms" make you feel smart? I agree with your point by why not just say it like a normal person? It's the most important trend in being able to occupy areas that are currently in long term droughts.
@Ampreon25@GinoTheGhost@hofman_eth I'm genuinely curious, should any of these things result in death? Sure, she did some things that are illegal, but we have a process for that. Giving one individual the power to end another persons life should come with extreme restraint. We have laws and sentencing guidelines.